3075
Proton Magnetic Resonance Spectra of Cubane Derivatives. I. Syntheses and Spectra of Mono- and 1,4-Disubstituted Cubanes John T. Edward, Patrick G. Farrell," and Gordon E. Langford Contribution from the Department of Chemistry, McCill University, Montreal. Canada, H3C 3C1. Received August 27, 1975
Abstract: The syntheses of a number of monosubstituted and 1,4-disubstituted cubanes are described, together with the measurement and analysis of their 100-MHz ' H N M R spectra. Typical coupling constants observed are 5.3 H z (vicinal), 2.5 Hz (four bond), and -0.7 H z (five bond). A simple additivity rule is described whereby chemical shifts in CDCI, can be predicted to within d~0.02ppm. Both chemical shifts and coupling constants are shown to var.y with substituent electronegativity. The derived correlations allow quick and effective identification of cubane derivatives from their ' H N M R spectra, and also aid in the interpretation of the more complex spectra of less symmetrical cage molecules.
Since the first syntheses of cubane' and some of its derivatives were reported more than 10 years ago by Eaton and C ~ l e , several ~ . ~ reports of synthetic and mechanistic studies involving these most interesting cage compounds have ap~ e a r e d . ~Derivatives -'~ of compounds of fixed, known geometry are of especial value for physicochemical studies of intramolecular group interactions because potentially complicating geometrical/conformational variations are minimized or eliminated. This was realized many years ago by Roberts and Moreland15 in their study of the influence of 4 substituents upon the acidity of bicyclo[2.2.2]octanecarboxylicacid, and since then many other workers have utilized derivatives of rigid alicyclic hydrocarbons in studies of polar effects.16 As the symmetry and rigidity of the cage make cubane derivatives excellent models for such studies, it is surprising that only one investigation of polar effects in this system has been reported.' As part of our own work on the mechanism of the transmission of electronic effects within molecules we had also chosen CUbane derivatives as models and report here details of the syntheses of a number of mono- and 1,4-disubstituted compounds. 1,4-Disubstituted derivatives of cubane are also of interest from the standpoint of their 'H N M R spectra, as they provide experimental examples of the fairly rare [AB]3I7system. Although the theory to account for the spectra of this system was worked out a number of years ago,l8 we are aware of few other nondegenerate examples of such s p e ~ t r a .The ' ~ 60-MHz IH N M R spectra of a number of 1,4-disubstituted cubanes have been reported, without analysis, but most show only a singlet absorption for the cage protons, even when the two substituents a r e quite different. This surprising simplicity has sometimes been attributed to approximately identical shielding constants for the two substituents.I2 However, as we show below, such deceptively simple spectra arise whenever the A and B chemical shifts differ by less than 5 Hz, and this situation can arise even with two substituents having very different shielding effects, or electronegativities. W e have recorded the ' H N M R spectra of cubane and 26 of its monosubstituted and 1,4-disubstituted derivatives and report here the first analyses of such spectra. The effect of substituents on chemical shifts in CDCI2 solution can be expressed in terms of additivity parameters which predict all observed chemical shifts to within f 2 H z a t 100 M H z . Three-bond, four-bond, and five-bond couplings are observed whose magnitudes show small but consistent variations with substituent electronegativity. Our analysis permits prediction of the ' H N M R spectra of 1,2- and 1,3-disubstituted cubanes, and for the known cubane- 1,3-dicarboxylic acid6 excellent agreement obtains between the predicted and reported spectrum. Edward, Farrell, Langford
Results Analysis of the Spectra. In all mono- or disubstituted cubanes, three geometries of proton spin-spin coupling are possible. Following Cole's designation2Qof cubane substitution patterns, we have termed these couplings oftho (along an edge of the cube), meta (across a face of the cube), and para (along a diagonal of the cube). The 1,4-disubstituted cubanes (I) form Ha Hn
I JAB JAA,, J'AB
= JORTHO JBBT = J M E T A = JPARA
an [AB13 spin system and the observed chemical shifts of HA and H Bdiffer by < I ppm, depending upon the substituents X and Y. The monosubstituted cubanes (11) form a much more complex [AB]3C system, and in all the cases we have studied the B and C chemical shifts are very close (less than 0.05 ppm
I1 JAW
=
JR,.
J A A ' ~ JRH',
J'AH
=
JOKTHO
JAC
=
JMWA
JiiAnA
difference). Thus their spectra are considerably more difficult to analyze than those of the 1,4-disubstituted cubanes. All spectra were analyzed using the LAOCN321 computer program and the results plotted on a Calcomp Plotter. Approximate values of coupling constants were estimated by first-order analysis of the nearly [AX13 spectra of the substituted 4-methylcubanes. These estimates were varied until the spectra computed from them were similar to the experimental spectra, and then the fit was optimized by iterative calculation. It is impossible to analyze the spectra exactly because of the enormous number of unresolved lines (less than 20 peaks and shoulders can be assigned in the average spectrum). Thus it is necessary to assign the calculated lines to fit under an ex-
/ Proton Magnetic Resonance Spectra of Cubane Derivatives
3076 Table 1. Coupling Constants in Substituted Cubanes (I)
Meta
Ortho Solvent
Y
X
Br
COCl
Br C02CH3
CDCI3
5.57 5.64 5.62 5.47 5.22 5.32 5.05 5.20 (5.08)d 5.26 (5.01)~
cc14
c02-
CH2OH
“2
c02-
CH3 Br COOH
CO2H Hc Hc
a
JAB”
C6D6 CAN CDC13 D20/NaOD CDCI, CDCI3 CDCI3
Para
JAA,~
JBBF~
J’AB~
2.85 2.85 2.74 2.41 2.44 2.62‘
3.17 3.08 3.37 2.93 2.36 2.84c 2.37 2.77 2.84
-1.36
LO3 2.35 (2.72)‘ 2.30 (2.43)e
- 1.20 -1.& -0.80 -0.69
-0.52 -0.35 -0.87 -0.67
In Hz, 1 0 . 1 Hz. In Hz, 1 0 . 2 Hz. The assignment of the A and B protons is not certain here. JBC ( i 0 . 3 Hz). e JAC ( 1 0 . 3 Hz).
perimental “envelope”, and more than one assignment will fit the same envelope. It was found that the iterative calculation would not consistently converge to the “best” fit (as judged by visual comparison of the experimental spectrum and the calculated plot), unless the trial values of coupling constants were very close to the optimal ones. Hence there is a probable uncertainty of f0.1 Hz in the calculated values of J A Band f 0 . 2 Hz in J A A ~J, B B and ~ , J’AB.Full analysis of coupling constants to this accuracy was possible only in well-resolved spectra -~g5 l 50 Hz. Table I shows the values of where IO 5 coupling constants found. Couplings were measured in a variety of solventsZ2for several reasons: to increase the solubility A u d , and to of some compounds, to induce variations in ~ U investigate the effect of solvent on coupling constants.
IUA
Typical computed and observed spectra are shown in Figures IUA - u d < 10 Hz the [AB13 multiplet collapses together rapidly and becomes a “singlet” for I U A u d 5 5 Hz. In this range, the observed spectra do not contain enough detail for an accurate measurement of the coupling constants; the computed spectra in Figure l a represent “typical” values of coupling constants ( J A B= 5.4 Hz, J A A=~ 2.7 Hz, J B B =~ 2.7 Hz, J’AB = -0.8 Hz). (This is possible since variations in coupling constants are small.) When the experimental [AB13 spectra were “singlets”, the half-width of the peak was used to calculate U A and U B . This measurement is limited by the resolution of the spectrometer when I U A - u d 5 2 Hz; in such cases the linewidth gives us an upper limit for 1 and 2. Note that for
IvA
-
Table 11. Chemical Shifts” in Substituted Cubanes (I) VB
VA
Y
Hc
CH3
Br
C02CH3
X
Hc COOH COOCH3 Br COCl COOH COOCH3 COCl CON3 NCO NHC02CH3 COOH COOCH3 COCl Br C02H COCl NCO NHC02CH3 CONH2 CH2OH CONHCONHRk NHCONHCOR~
COCl C02H CO3‘Bu CHzOH
C02CH3 COCl CO2H CO3‘Bu CHzOH
“2
c02-
“I+
COOH
Obsd
Calcdb
403.8 428.2 424.1 425.5 443.5 41 1.5 407.6 427.2 410.5 390.6 389.7 434.8 430.6 446.9 428.8 425.6 f I h 440.0 407.6h 407.3h 420.2h*i 387.5 430Sh*‘ 414.6 f 1.5h 422.2 443.6 428.7 433.8 379.6 369.9’” 421.4 i 1.7h,”
403.8 429.2 425.3 425.3 443.3 41 1.5 407.6 425.6 391.4 390.8 433.8 429.9 447.9 429.8 426.9 441 .O 406.8 406.2 388.1 423.0 444.7 428.6 379.1
Obsd c
401.9 400.2 409.5 404.9 364.4 363.1 368.6 363.5 349.9 349.8 426.2 423.7 424.9 c
425.6 1l h 425.6 41 2.6h 410.9h 423.2h,i 413.5 425.3h,i 414.6 11 . 5 h
Calcdb
Other Y
c
c 398.2d 397.7id 368.7‘ 407.4d 402.5d 127.u 127.0;f 368.3‘ 130.9
403.2 401.5 408.4 405.2 364.8 363.1 366.8 35 1.3 350.4 424.6 422.9 426.6
126.0‘ 126.0;f 365.6g 369.3‘
c
424.7 426.7 411.2 410.3 4 14.0
c c
c c
C
C
c c 384.8‘“ 421.4 f 1.7h,“
c c
370.0‘ 370.7e 368.8e 368.8;‘ 367.2g 369.5e 369.1;e 375.01’ 370.7e 369.7e 369.9‘ 133.3‘ 376.7J
“ Chemical shifts in Hz at 100 MHz. All samples in CDC13 except as noted. Shifts calculated by eq 1. Same as HA. Hc shift. e OCH3 shift. f CH3 shift. NHC02CH3 shift. Unresolved singlet for A and B protons; IYI - u d estimated from line width. The reverse assignment of A and B shifts is also possible. J C H 2 0 H shift, Ir R = (cubyl)C02CH3. C(CH3)3 shift. Solvent is D2O with TMT as external reference. Solvent is DZO/NaOD, with TMT as external reference. Journal of the American Chemical Society
/
98.11
/
May 26, 1976
3077
I
I
I
440
420
400
Hz.
Figure 2. The computed and observed [AB]3C spectrum of bromocubane.
Table 111. Substituent Additivity Constants (CDC13) 1
I
370
350
1
390
Hz.
C
Substituent
Aug, HZ
H C02CH3 CH3 Br COCl CO2H NCO NHC02CH3 CH2OH CON3 CONHz CONHCONHR‘ NHCONHCORC
!
lu,, Hz
0
+21.5 -38.4 +21.5 +39.5 +25.4 +5.3 +4.7 -13.4 +25.6 1 8.7b +29.1d +13.2 f 1.5
+
l u a , HZ
0
0
-2.3
-6.1
-17.7
a
+4.6 +1.4 -0.6 -14.1
C3.6 -1.3 -5.6
-15.0
a
-11.3
a
-0.8
a
-2.16 +O.ld
a a
-10.6 F 1.5
a
a
Not measured. Reversing the assignments of U A and V B would = -5.1. R = (cubyl)CO2CH3. Reversing the give AB = 21.7, assignments of U A and U B would give l g = 23.9, L., = 5.3.
Figure 1. Typical computed and observed (AB13 spectra of I ,4-disubstituted cubanes. (a) Computed spectra for I U A -.ug( < I O H z and coupling constants as indicated in the text. (b) Observed and calculated spectra for 4-bromocubane-I-carboxylic acid chloride in benzene solution, I V A - u d = 15.9 Hz. (c) Observed and calculated spectra for 4-methylcubane-lcarboxylic acid in chloroform solution, I V A = 37.3 Hz.
UBI
Figures 1 b and I C show observed and calculated spectra where the two meta coupling constants are not equal, J A A f # J B B , .Note that the “A” and “B” halves of the spectrum are mirror images when J A A J = J B B , ,but the spectrum becomes unmistakably asymmetrical when J A A , # JBB,. This feature of the spectra makes it very easy to observe any substituent effect upon the meta coupling constants. This characteristic of [AB13 spectra notably distinguishes them from [AB12 spectra, which a r e always syrnmetrical.lsb Assignment of A and B chemical shifts was done by several
Edward, Farrell, Langford
methods. I n monosubstituted cubanes the assignment is unambiguous because of the asymmetry of the system. I n 4methylcubane derivatives, the protons to the methyl group appear as a slightly broadened multiplet; we have shown by spin decoupling that this broadening is due to a small longrange coupling with the methyl group. I n other [AB13 spectra, the shifts were assigned on the basis of a substituent effect additivity rule, and of the observed substituent effects on coupling constants, which are discussed below. For systems with I I J A- v g ( 5 5 H z it was sometimes not possible to definitively assign the shifts; in such cases the shifts are quoted as Vav f A and vav T 3 where 3. is I V A - 4 . / 2 , The Chemical Shifts. Chemical shifts in CDC13 (or DzO) of all compounds studied are listed in Table 11. Where literature values are available, they are generally in good agreement with 3ur measurements. One of the largest discrepancies is the value for cubane itself, reported as 4.00 ~ p m but , ~which we find as 4.038 ppm (403.8 Hz at 100 MHz). The observed chemical shifts in CDC13 of each cage proton in the compounds of Table I1 can be predicted within 2 H z from eq 1 by addition of all the appropriate substituent shift constants AIJ,shown in Table 111, to the chemical shift of cubane itself. VCDC13
= 403.8 -I-
i
b ( i ) (HZ)
(1)
Figure 3 illustrates the application of this equation. For completeness, we include a summary of chemical shifts that have been reported (in CDCI3) for other cubane derivatives (Table
/ Proton Magnetic RFsonance Spectra of Cubane Deriuatiues
3078
W02C
anc L
I
= 403.8
= 4 0 3 . 8 + AW.,IC02Me)
vC = 4 0 3 . 8
'A
+ svglco2Me) = 425.3
Table IV. Other N M R Data from the Literatureo (CDC13)
HP
= 4 0 i . 5 Hz
+ AvglCO2Me) = 3 9 7 . 7
4.12,bs
Y A =: 410.3, U B = 413.1
CHzOH
NHC02CH3 HC
3.85,c br s
YA
2"
4.1 5 , b s 4.05,bs 4.20,bs 3.95-4.43,b m 4.32,d s 4.00,c br s 4.02,c m 4.27,cs
"B
vA
Br
yB =
403.8
+
AY
403.8
f
A v g I B e ) + Av l C 0 2 M e ) = 4 2 2 . 9 H I
lco2Me)
+
AY
IBr) = 4 2 9 . 9
Br Br Br Br CN CH20-PNB OAc CO3'Bu
Figure 3. A typical determination of chemical shifts using eq I
IV) and the additional substituent shift constants that can be estimated from these data (Table V ) . Discussion Syntheses. Even though the syntheses of several of the cubane derivatives used in this study have been reported by other ~ ~ r k e r ~ we , ~ consider ~ . ~that ~ some - ~ brief ~ - comment ~ ~ upon the overall synthetic scheme (Scheme I ) is of value, in view of the considerable current interest in cage compounds. Compound 1 can be made by either the original method of Eaton and Cole, or one of its modifications, e.g., that of Chapman et a1.I0 Conversion of 1 to 21 is straightforward, and we have followed the procedure of Luh and Stock" to obtain good yields of the diacid 22. It is very important that the reaction mixture be kept ice-cold when it is neutralized with HCI during the workup. I n the Favorskii reaction, Cole20 obtained side products arising from reduction to the corresponding homocuban-9-01, W e have also observed this reduction with com-
-
All spectra measured at 60 MHz; shifts in ppm from TMS. See ref 24. See ref 20. Spectra measured in CDCIl or CCId. See ref 9.
Table V. Estimated Substituent Additivity Constantso from Data in Table IV Avo, H Z
1
2,X=H
i
3, X = Br 4, X = CH,
Po \-
2"
OH CI OAc CH2OPNB
21
3 7
7
4 (36)
-20 * 4 -574 -23b or + I -4 F 4 (-8')
~~
I n H z at 100 M H z to conform with Table 111. Most probable values (see Discussion).
fly
@co2H-
X
X 8,X=H
5,X=H
9, X = Br 10, X = CH,
.6, X = Br 7, X = C H ,
11, X = H; Y = C0,Me
12, X = H ; Y = COCl 13, X = H, Y = Br 14, X = Br; Y = C0,Me 15, X = Br; Y = COCl 16, X = CH,; Y = COCl 17, X = CH,; Y = C0,Me 18, X = CH,; Y = CON,
19, X = CH,; Y = NCO 20, X = CH,; Y = NHC0,Me
flC02H
----.t
@C02Me ----t
X 30, X = COLMe 31, X = COCl 32, X = C0,'Bu 33, X = H 34, X = Br 35, X = CH,OH
* 4 (Ob)
-10 F 4
*
CO,H
~
Au?, H z
27 f 4 (30b) 31 & 4 (35b) I f 4 -3f4 12&4 18b or -6 -4 4 (O b)
C03'Bu CN
X
X
OH CI OAc C02CH3 Hc HC C02CH3
Po -
Scheme I
(s) = 390.4, U B = 392.5 (s)
HE
HA
COlH
Predicted from Table I11
Y
Br
Ma02C
Reported spectrum
x
HZ
CO,H
X
CO,H 22
Journal of the American Chemical Society
23
24, R = CH,; X = 25, R = CH,; X = 26, R = CHj; X z= 27, R = CH,; X = 28, R = C H j X =
COCl CH,OH CONH, NHC0,Me
NCO 29, R = H X = NH,+C1-
/
98:ll
/
May 26, 1976
3079 pounds 5 and 6, in 25% aqueous N a O H a t 1 15 OC, isolating about 10% yields of 5a and 6a, respectively. However, with compounds 6 or 7 in 25%aqueous KOH, yields of the 9-01 were
dilute D20-NaOD showed about 50% decomposition (by NMR) after 24 h, and a reddish-brown precipitate formed. N o decomposition products were identified, but the N M R spectrum of the mixture showed resonances in the alkene region. As an alternative approach to the synthesis of the amino acid, we attempted a Hofmann rearrangement from the ester amide 26. However, the anticipated product, urethane (27), was isolated in very poor yield, the major product being the acylurea 38 which was apparently formed by addition of
@ P:H
X
X 5a, X
=
H
X
=
Br
5,X=H
6, X = Br 7, X = CH,,
6a,
7 4 X = CH,
small or negligible. T h e mechanism of this reduction remains to be explained. The isolation of dicarbinol35 by the direct reduction of diacid 22 with lithium aluminum hydride proved difficult because of the facile isomerizationI3of the dicarbinol to homo-
fl HH2 E@ CH,OH
35
22
+
OH @OH
CH,OH
37
36
cubane and bishomocubane derivatives (36,37). However, in the course of our synthesis of the carbinol ester 25, we found that 35 could be obtained without rearrangement by borohydride reduction of ester acid chloride 24 or even diester 30 a t gcoclNaB& temperature: room ~
SaB& __t
__t
8h
10 min
C02Me
MeO;C
28
26
0
0
II
C-N-C-N
II
+
LiA,H, -c
COOH
C6,Me
C0,Me 24
25
pf q g f HOCH,
CH20H
MeO&
C0,Me 35
30
This reaction is somewhat surprising because esters are not normally so easily reduced by sodium borohydride. Furthermore, under the same conditions (NaBH4, dioxane, 12 h, room temperature) the bromocubane ester 1 4 gave no reaction. A Curtius rearrangement of the azide derived from the half-ester 23 was used to synthesize 4-aminocubanecarboxylic acid, isolated as the hydrochloride 29. Like other cubylamines," the sodium salt of 29 was unstable; a solution of 29 i n
Edward, Farrell, Langford
\
/
MeO,C
COLMe 38
starting material 26 to the intermediate isocyanate 28. Similar acylureas have been previously reported as products of Hofmann rearrangement^,^' but this coupling reaction is normally significant only when N a O H is the base, and when the amide is present in excess. Coupling Constants. From Table I, it can be seen that the vicinal coupling constants ( J o r t h o ) are consistently ca. 5.3 Hz, ca. 2.5 Hz, and five-bond coupling four-bond couplings (Jmeta) ca. -0.7 Hz. Typical cis vicinal couplings in constants (Jpara) cyclobutanesZ8 are approximately 10 Hz, or about twice the cubane value. If no other effects were operative, the Karplus would predict that the vicinal coupling in cubane, with zero dihedral angle, should be larger than that in cyclobutanes where, because of puckering of the ring,30 the average dihedral angle is greater than zero. This apparent discrepancy is explained by the much greater strain of the cubane skeleton. Theory predicts that a vicinal coupling constant J A B should decrease as the bond angles HA-C-C' and C-C'-HB are in~ r e a s e d , and ~ ' experimental evidence supports this view.'* I n cubane, the H-C-C' bond angles are all ca. 125' vs. ca. 109.5' for unstrained aliphatic hydrocarbons and ca. 113' 30c for ordinary cyclobutanes. Strained cyclobutane rings in other fused-ring structures, with H-C-C' bond angles greater than 1 13", also have small vicinal coupling constants.28 It is not surprising that the meta coupling constants observed in cubanes are larger than the cis cross-ring couplings reported in most ordinary cyclobutanes.28Many examples have been reported32 of abnormally large four-bond couplings in strained saturated systems, especially when the protons in question and the three carbon atoms joining them lie on a roughly planar "W" path. Examination of the cubane skeleton shows that such W pathways link the meta protons and that for each meta interaction there are two equivalent pathways possible. The positive signs observed for these couplings in cubane are in accord with theory.32b The nonzero, negative, five-bond couplings which we have observed between para-situated protons in cubane are particularly interesting. Relatively few five-bond couplings have been reported in aliphatic systems; these generally involve protons
1 Proton Magnetic Resonance Spectra of Cubane Dericatices
3080
Coupling Constants. Solvent Effects and Substituent Effects.
0.0
XU,
0.0
I .o
0.5 (Toft)
0.5
1.0
r,( T o f t )
Because of the rigidity of the cubane skeleton, bond lengths and dihedral angles are independent of the s ~ b s t i t u e n tThus, .~~ the usual conformational variations of coupling c o n ~ t a n t s ~ ~ . ~ ’ will be absent in all cubane derivatives. Steric effects should be insignificant for normal-sized substituents. Hence, any variations in coupling constants would be expected to arise from electronic effects of the substituents, or from solvent effects. When the spectra of a given compound in various solvents were analyzed, coupling constants were found to be constant, within experimental error (Table I). Solvent effects on couplings thus appear to be insignificant in the cubane system. This result is in accord with previous observations. Although solvent effects on geminal coupling constants are well documented, there is no unequivocal evidence for solvent effects on vicinal or long-range coupling constants in conformationally rigid systems.35 On the other hand, our studies show clear evidence that couplings of all three types in cubanes are influenced by the electronegativities of the substituents. The effects, although small, are unmistakable. Substituent-induced differences are easiest to detect between the two meta couplings in a single substituted cubane because they impart a distinctive asymmetry to the spectrum: the “A” and “B” multiplets cease to be mirror images (e.g., see Figures 1 b a n d IC). The results in Table I indicate that the meta coupling constant tends to be smaller between the protons to the more electronegative substituent. T o show more clearly the effects of substituent electronegativities on the coupling constants, we have plotted the coupling constants from Table I vs. the sum of the electronegativities of the 1 and 4 substituents. Relatively similar correlations were obtained using a variety of measures of electronegativity: E R ,and ~ ~“mutually consistent group electronegativi t i e ~ ” In . ~Figure ~ 4 we show how the ortho, meta, and para couplings vary with the sum of the “weakly protonic solvent” 01 values of Taft36 for the 1 and 4 substituents. In this figure, the same 01 value was used for COO- as for C O O H and C O O R (0.21) since these substituents are thought to have similar e l e c t r ~ n e g a t i v i t i e s .Since ~ ~ there are two meta couplings in a 1,4-disubstituted cubane, we have plotted their average value, (JAA, JBB’)/2. It is clear from Figure 4 that the ortho coupling constant JAB increases with the electronegativity of the substituents, while appears to decrease (Le., becomes more the para coupling J’AB negative). There is little clear trend in the average meta coupling constants; if anything they appear to increase slightly with electronegativity. There have been a number of studies of electronegativity effects on vicinal coupling constants in aliphatic molecules.40 When the substituent is a to one of the coupled protons, as in substituted ethanes, it is well established that the vicinal coupling constant decreases with increasing electronegativity. However, several ~ t u d i e s ~suggest ’ . ~ ~ that when the substituent is one bond further removed from the coupled protons, as it is in cubane, the effect will be reversed, and J,,, will increase with substituent electronegativity, as we have observed. Considerably less is known about electronegativity effects on long-range couplings in aliphatic molecules. The four-bond coupling constants in 2,2-disubstituted propanes have been reported to increase with substituent e l e ~ t r o n e g a t i v i t y . ~ ~ ~ Conversely, electronegative substituents appear to cause a decrease in the four-bond coupling constant in mono- and 1 , l -disubstituted acetones.43b In both cases the effects were small and the molecules studied were not conformationally rigid, so that no firm conclusions can be drawn. Our own results for meta couplings are also equivocal. The smaller meta COUpling constants are observed between the protons near the more ~
1
,
~
~
s
~
’
+
I 0.0
I
0.5
1
I I .o
(Toft)
Figure 4. The variation of coupling constants with Taft UI values for weakly protonic solvents: (a) ortho couplings, (b) meta couplings, (c) para cou-
plings.
joined by planar multiple zig-zag paths, and are reported to be positive.32 However, in most of these systems, the sign of the coupling constant has only small effects upon the overall spectral appearance and unequivocal assignment of signs is difficult or impossible. The cubane system is ideal for such studies because its rigid geometry makes long-range couplings possible while the symmetry allows detailed analysis of spectra, and in all cases studied here the five-bond couplings were negative, as would be expected if a “through space” mechanism is involved.33 Although the C I - C ~ separation (2.7 A) in cubanes is somewhat greater than the usual 2.2 A range for such couplings,33 the atoms are aligned perfectly for back-lobe overlap. Alternatively, through-bond interactions between the para positions should be enhanced by the presence of six inductive pathways between C I and C4. Journal of the American Chemical Society
/ 98:ll / M a y 26, 1976
308 1 electronegative substituent (Table I), but the average meta coupling constants appear to increase slightly with electronegativity. These two observations can be reconciled by the hypothesis that an electronegative substituent causes a decrease in the value of the nearer meta couplings, and an increase in the more remote ones. Vicinal coupling constants provide a precedent for such a n alternation of effects. W e are not aware of any other studies of substituent effects on five-bond couplings in rigid aliphatic molecules. In substituted benzenes, para coupling constants have been reported to decrease with increasing electronegativity of the substitue n t ~This . ~ trend ~ agrees with our observations for cubane, but the agreement may not be significant, because of the very different geometry and electronic hybridization of benzene. Chemical Shifts. Unsubstituted cubane has a chemical shift of 4.04 ppm, a remarkably large shift for an alicyclic compound, but explained by the high degree of s character of the C-H bonds (ca. 319/oZ0).For the substituents we have studied, chemical shifts vary over a range of about 0.8 ppm a t the /3 position and about 0.25 ppm at the y position. Thus the /3 and y effects are comparable in magnitude to those reported in other cage s y ~ t e m sand ~ ~even . ~ ~alicyclic compounds.29bOne would expect substituent effects to die off with distance; however, for the few compounds for which we have data, the y and 6 effects are comparable in magnitude. If this trend is confirmed for a wider range of substituents, it may be good evidence in support of an inductive mechanism for these substituent shift effects. The cubane skeleton provides six different paths along C-C bonds between a substituent and the 6 proton; thus the inductive theory predicts an unusually large 6 effect. l 6 (There are only two significant inductive paths to the y substituent, and one to the /3 substituent.) However, none of the through-bond pathways in cubane possess the parallel alignment with the C-H and C-X bond orbitals that is considered favorable for inductive interaction^.^^,^^ It is also possible that this large 6 effect arises as a result of direct interaction via orbital overlap “within” the cage, Le., a through-space effect, made possible by virtue of the geometry. It is not possible to distinguish between these possibilities from the data presented. In Figure 5 the relationships between substituent effects at the /3 and y positions a r e shown. There is an excellent linear correlation (Figure 5a) between the /3 and y effects of all carbon-linked substituents that have been accurately measured. Substituent effects known less accurately (generally calculated from literature spectra) are plotted to show all linear combinations of /3 and y effects which fit the available data. For carbon-linked substituents whose chemical shifts are not known accurately, we have used the linear correlation in Figure 5a to assign “most probable values” of /3 and y effects, and these are included in Figure 5c. As Figure 5b shows, there is no general correlation that will fit /3 and y effects for all substituents, but it remains possible that separate linear correlations may exist for nitrogen-linked, oxygen-linked, and halogen substituents. Figure 5c illustrates a simple graphical method of calculating I U A - v d for any pair of substituents, and also shows why so many 1,4-disubstitutedcubane derivatives reportedly give “singlet” spectra a t 60 MHz. For the purpose of illustration of the method the carbomethoxy group is chosen as one substituent and a line of unit slope drawn through this point. For any second substituent, “ Y ” ,on the cage it can easily be seen that the value of I v~ - v d is given by the difference along the abscissa between the line of unit slope and the position of Y on the graph (Figure 5c). As was pointed out above (see Figure l a ) , for values of I v A - UBI 5 5 H z (-0.08 ppm a t 60 M H z ) “singlet” spectra are obtained. Thus, for all substituents “Y” within the region of Figure 5c bounded by the dotted lines, the spectra of the 4-Y-carbomethoxycubanes will appear as singlets a t 60 M H z . It so happens that the majority of combinaEdward, Farrell. Langford
I
I
-40
I -20
I 0
I 20
I 40
0
0
$0 N1
-40
- 20
0
20
40
8‘
E:
IC
0.
2” -I 0
ru
-
0
/
I 1
- 40
-
I 20
I 0
1 20
I 40
4 Figure 5. The relationships between substituent effects upon the chemical shifts of protons at the /3 and y positions. (a) Carbon-linked substituents. oxygen-linked substituents; ( A ) (b) (0)Nitrogen-linked substituents; (0) hydrogen and halogen substituents. (c) A graphical illustration of the conditions leading to singlet spectra at 60 MHz.
tions of substituents used in previously reported studies of cubanes lie within this region of Figure 5c and hence “singlet” spectra were recorded. Note that a “singlet” does not imply that the X and Y substituents necessarily have similar shielding effects. Many pairs of substituents, e.g., Br and O H , or C 0 2 M e and NHC02CH3, have very different effects a t the fi and y positions individually, but still give rise to singlet spectra at 60 M H z or even 100 M H z . If the chemical shift effects of substituents arise from through-bond interactions, as has been discussed above, then such effects should be related to some measure of substituent electronegativity. W e have chosen to correlate the data obtained here with the UI values of Taft, for weakly protonic solvent,36 as shown in Figure 6, because these values should be most applicable to the cubane system in CDCI3. Relatively
/
Proton Magnetic Resonance Spectra of Cubane Deriuatiues
3082 a
l
1
a 0
P
0
9
a
0
d
I
450 b
.o 2
0
0 2
0 4
,
I
I
425
400 Hz.
06
U,
0
0
\ -0.2
I
1
I
0
0.2
0.4
Figure 7. Predicted proton spectra at 100 MHz for: (a) cubane-1.3-di= 5.4 Hz; carboxylic acid, (b) cubane-1.2-dicarboxylic acid. JORTHO JMETA = 2.7 Hz; JPARA I :-0.8 Hz.
I 0.6
Ul
1-0' 2
I
0
I
0.2
I
I 0.6
04
=I
Figure 6. The variation of chemical shifts with Taft 61 values for weakly protonic solvents: (a) 0-hydrogen atoms, (b) y-hydrogen atoms, (c) 6-hydrogen atoms.
similar correlations obtain with other a1 scales,37 but the chemical shift effects correlate only poorly with group electr~negativities.~~ Although there is considerable scatter, the correlation is as good as is normally observed in other systems.40The slopes of plots for the b, y,and 6 chemical shifts against electronegativity all have the same sign, in sharp contrast to the I-adamantyl system, where the slopes were found to alternate in sign.45 Again the correlation is particularly good if only carbon-linked substituents a r e considered; only the CN group deviates appreciably. In their similar correlation of data for adamantane derivatives Schleyer and Fort observed that halogen substituents required a separate line45and it has been observed that hydrogen frequently deviates from a1 correlation^.^^ Such dependence on the atom Z through which the substituent is linked may be attributable to anisotropy of the C - Z bond, but we do not have sufficient data a t present to confirm this suggestion. If anisotropy is important in such cases, it is of interest to Journal of the American Chemical Society
/
98.1 1
/
examine why carbonyl substituents (which a r e certainly anisotropic) correlate well with other carbon-linked substituents. Using Pople's calculations of carbonyl anisotropy effects,29c and averaging these effects over all possible conformations of the carbonyl group in cubanecarbonyl compounds, we estimate the carbonyl anisotropy effects to be less than 0.01 ppm at both and y positions. Hence these small effects will not significantly perturb the relationship found for carbon-linked substituents. I n general, Figure 6 suggests that polar effects a r e more important than magnetic anisotropies in determining the net chemical shift effect of a substituent on the cage protons of cubane. The remarkably precise additivity (f0.015 ppm) observed in this work for substituent effects on chemical shifts is worthy of note. Many additivity correlations have been reported prev i o ~ s l ye.g., , ~ ~ for substituted methanes, ethylenes, and cage molecules such as adamantanes, but few of these are accurate to even f O . l ppm. Probably the extreme rigidity of the cubane cage is one reason for our success, together with the facts that (a) all substituent effects i n the cubanes a r e relatively small (I Ad < 0.5 ppm for all substituents studied) and (b) the only polysubstituted cubanes studied so far are the 1,4-disubstituted ones where the distance between the substituents is greatest, and thus their interactions are least. Applications of Our Results. For any combination of the substituents discussed in this paper, spectra can be predicted with considerable accuracy, and for other substituents approximate predictions can be made on the basis of their electronegativities. Furthermore, the N M R parameters which we have measured for mono- and l ,4-disubstituted cubanes can be used to predict spectra for other cubane substitution patterns. Figures 7a and 7b show the spectra which we predict for cubane1,3-dicarboxylic acid and cubane- 1,2-dicarboxylic acid, respectively, using typical coupling constants from Table I and chemical shifts calculated from eq 1. The 1,3-diacid has been reported6 to show a broad absorption in the NMR between 6 3.90 and 4.70 ppm (presumably a t 60 MHz), in excellent agreement with Figure 7a. The synthesis of the 1,2-diacid has
May 26, 1976
3083 not yet, to our knowledge, been reported. T h e substituent effects on coupling constants for these substitution patterns can only be estimated from Table I and Figure 4 but such variations should be small. Finally, the chemical shift effects and coupling constants observed in the cubane system are useful in analyzing the much more complex spectra of less symmetrical cage molecules, e.g., homocubanes, bishomocubanes, and b a ~ k e t a n e s .In~ ~analyzing any complex spectrum it is necessary to estimate a good set of approximate chemical shifts and coupling constants before attempting to derive accurate values by an iterative calculation. The coupling constants shown in Table I and their variations with substituent, together with the effects of substituents on chemical shifts shown in Tables 111 and V, are very useful in making such estimates. W e are currently using these data in the analysis of the spectra of a number of homocubane derivatives, which we expect will clarify the influences of strain and geometry on spectral parameters.
Experimental Section N M R spectra were measured on a Varian HA-I00 spectrometer a t 30 OC using field sweep. Tetramethylsilane (10%) was added to samples as internal reference and lock, except for spectra in DzO, where an external reference and lock, tetramethyltin, was used. The chart frequency was calibrated a t 50-Hz intervals (maximum deviation < I .5 Hz). Reported values are thus considered accurate to & 1.5 Hz and reproducible to f 0 . 5 Hz. Melting points were determined in a Gallenkamp micromelting point apparatus in sealed capillary tubes and are uncorrected. Infrared spectra were recorded on a Perkin-Elmer Model 257 spectrometer. All compounds synthesized were assayed by mass spectrometry and gave satisfactory spectra.50 1-Bromopentacyclo[ 4.3.0.02,5.03~8.04~7]nonan-9-one 4-carboxylic ethyleneketalacid(1)wasprepared by the method of Chapman et a1.,I0 m p 191-192.5 OC (lit.3*'0187-189 "C). Pentacyclo[4.2.0.02~5.03~8.04~7]octanecarboxylic acid (8) was prepared from 1 by the method of Eaton and Pyrolysis of the tert-butyl perester of 1 in boiling cumenegave 2, mp 66-67 OC (lit.3 64-65 "C), which was deketalized in 75% w/w HzS04 to 5, mp 93-94 OC ( l k 3 90-91 "C). Heating 5 in 25% aqueous N a O H for 2 h a t 1 1 5 OC yielded the cubane acid 8,mp 122.5-124 O C (aqueous MeOH) (lit. 124-125 0C,20125-126 oC7b),together with small amounts (ca. 10%) of 1-bromopentacyclo[4.3.O.Oz~5.03~8.04~7]nonan-9-oI (5a): mp 80-83 OC (lit.20 84-86 "C); N M R (CDC13) 6 4.18 ( 1 H , d, J = 2.0 Hz, C H - O H ) , 3.83-3.43 (4 H, m), 3.43-3.26 (2 H, m), 3.26-3.09 ( 1 H, m), 2.03 ( I H , broad s, O H ) . 4-Bromopentacyclo[4.2.O.O2~5.O3~8.04~7]octanecarboxylicacid (9) was prepared from 1 as described by Klunder and Zwanenburg.I2 A Hunsdiecker reaction of 1 gave 3, mp 141.5-143 OC (lit. 143-144 "C,I2 138-141 O C Z 3 ) ; deketalization of this in 75% w/w H2SO4 gave 6, m p 143-144 OC (lit. 143-144,12 132-134 OCZ3),which was converted, on heating for 4 h in 25% aqueous K O H a t 115 OC, to the 4bromocubanecarboxylic acid (9), mp 213.5-216.5 OC (lit.7b210 OC dec). Treatment of 6 with 25% aqueous N a O H at 1 15 OC for 4 h gave 9 plus I,4-dibromopentacyclo[4.3.O.O2~5.O3~x.O4~7]nonan-9-o1 (6a) (10%):mp 116-1 17 OC (lit.24 116-1 16.5 "C); N M R (CDC13) d 4.20 ( I H , d , J = 2.1 Hz, CHOH),4.06-3.48 (5 H , m ) , 3.48-3.27 ( 1 H, m), 2.07 ( 1 H, broad s, O H ) . 4-Methylpentacyclo[4.2.O.O*~5.O3~8.O4~7]octanecarboxylic Acid (10). A solution of 1 (10.91 g, 0.0365 mol) in dry T H F (250 ml) was refluxed overnight with lithium aluminum hydride (1.25 g, 0.033 mol) under nitrogen. After cooling and addition of ethyl acetate (5 ml) and 10% sulfuric acid (300 ml), the THF was distilled off and the aqueous residue extracted with chloroform (3 X 150 ml). The chloroform extracts were washed with 5% aqueous sodium bicarbonate (2 X 100 ml) and water (2 X 100 ml), then dried (MgS04) and evaporated to the crude l-bromo-4-hydroxymethylpentacyclo[4.3.O.O~5.O3~s,O4~7]nonan9-one ethylene ketal (39)which on recrystallization (hexane) gave 8.78 g (84%) of colorless needles, m p 86.5-87.5 O C (lit. 80-84,23 84-87 "C5').Alcohol 39 was converted in the usual way to l-bromo-4tosy lox y methy lpentacyclo[4.3.0.02~s.03~8.04~7]nonan-9-one ethylene ketal (40), mp 115-1 17 OC (lit.23 117-1 18.5 "C). A solution of 40 ( 1 0.3 g, 0.0234 mol) in dry ether (450 ml) was refluxed with lithium
-
Edward, Farrell, Langford
aluminum hydride (4.6 g, 0.12 mol) for 24 h under nitrogen. The mixture was cooled and ethyl acetate (-20 ml) was added to destroy excess hydride. A solution of 10% v/v H2S04 (500 ml) was then added; the mixture was stirred for 30 min and separated, and the aqueous phase was washed with ether (3 X 300 ml). The combined organic phases were washed with 5% aqueous sodium bicarbonate (2 X 80 ml) and water (100 ml) and dried (MgS04). Evaporation and recrystallization (methanol-water) gave 4.6 g (74%) of l-bromo-4methylpentacyclo[4.3.O.O2~s.O3~8.O4~7]nonan-9-one ethylene ketal (4): mp 98-99.5 "C (lit.2382-84 "C), N M R (CDCI3) d 4.41-3.81 (4 H, sym [AB12 m, ketal H ) , 3.58-3.36 (2 H , m), 3.32-3.09 (3 H, m). 2.89-2.71 ( 1 H , m ) , l.l5(3H,s,CH3).Theketal4(4.35g,O.O162 mol) was stirred for 5 days in 75% w/w sulfuric acid (280 ml). The black mixture was then poured onto ice (300 ml), filtered with Celite, and extracted with chloroform. The extracts were washed with 5% aqueous sodium bicarbonate ( 1 50 ml) and evaporated to the crude 1 -bromo-4-methylpentacyclo[4.3.O.Oz~5.O3~8.O4~7]nonan-9-one (7),3.3 g (91%) which was converted directly to 10 without purification. A solution of 7 (3.3 g, 0.0147 mol) in 25% w/w aqueous potassium hydroxide (100 ml) was heated for 2 h at 115 OC, cooled, and washed with chloroform. The aqueous phase was acidified to pH 3 by dropwise addition of concentrated hydrochloric acid with stirring in an ice bath, keeping the temperature of the solution below 5 OC. The solution was extracted with chloroform (3 X 100 ml) and the extracts dried (MgS04) and evaporated to a brown solid which on extraction with hexanes gave I .29 g (54%) of 4-methyIpentacycl0[4.2.0.0~~~.03~*.04~7]octanecarboxylic acid (lo), mp (methanol-water) 142.5143.5 "C (lit.I2 139.5-141.0 "C). Pentacyclo[4.2.0.02~s.03~8.04~7]octanecarboxylic Acid Methyl Ester (11). A solution of 8 (0.130 g, 0.00088mol) i n methanol ( 1 5 ml) was refluxed with stirring over beads (50 mg) of Rexyn 101 ( H ) (a strongly acidic ion exchange resin) for 12 h. Filtration and evaporation gave the ester 11 as an oil which solidified on standing and was purified by sublimation. Yield 0.094 g (66%) colorless crystals, mp 5 1.8-52.5 O C . Pentacyclo[4.2.0.02~s.03+8.04~7]~ctanecarboxylic Acid Chloride (12). A solution o f 8 (0.041 g, 0.00028 mol) in thionyl chloride (4 ml) was refluxed for 4 h and then evaporated to an oil which was pure I2 by ' H N M R , but hydrolyzed rapidly back to 8 on contact with air. Bromopentacyclo[4.2.0.0z~s.03~8.04~7]o~tane (13). A solution of bromine (0.034 g, 0.00021 mol) in dibromomethane (2 ml) was added dropwise with stirring over 15 min to a refluxing mixture of 8 (0.025 g, 0.00017 mol) and mercuric oxide (0.025 g, 0.00012 mol) in dibromomethane (10 ml). After 3 h the mixture was cooled, filtered, and evaporated. The yellow residue was extracted with hot pentane and the extracts chromatographed with pentane on a silica column. Evaporation of the eluate gave 13 as a mobile oil, pure by ' H N M R . 4-Bromopentacyclo[4.2.O.~~5.O3~~.~~7~ctanecar~xylic acid methyl ester (14)was prepared from 9 as described above for 1 1 . Recrystallization (methanol-water) gave 14 as white flakes. mp 120.5- 122 "C (Iit.I2 119-121 "C). 4-Bromopentacyclo[4.2.O.O2~s.O3~8.O4~7]0ctanecarboxylicacid chloride (15), prepared from 9 as described above for 12, was obtained as an oil, pure by ' H N M R , which hydrolyzed readily back to 9 on contact with air. 4-Methylpentacyclo[4.2.O.O2~5.03~8.O4~7]octanecarboxylic acid chloride (16), prepared from 10 as described above for 12, was obtained as an oil, pure by I H N M R . It hydrolyzed rapidly back to 10 on contact with air. 4-Methylpentacyclo[4.2.O.O2~5.O3~8.04~7]octanecarboxylic Acid Methyl Ester (17).The acid chloride 16 was dissolved in methanol ( I ml) at room temperature. Evaporation gave 17 as an oil which was pure by ' H N M R . Curtius Rearrangement of the Azide Derived from the Methyl Acid 10. A solution of triethylamine (0.105 g, 0.001 04 mol) in acetone (2 ml) was added dropwise to a stirred solution of 10 (0.152 g, 0.00093 mol) in acetone (2 ml) and water (0.2 ml) at 0 "C. A solution of ethyl chloroformate (0.1 13 g, 0.001 04 mol) in acetone ( 1 ml) was then added dropwise over 30 min, and after a further 30 min at 0 O C sodium azide (0.085 g, 0.0013 mol) in water (0.4 ml) was added. After 2 h, the mixture was poured onto crushed ice, extracted with benzene (3 X 70 ml), dried, and evaporated at 20 OC. The resulting sticky yellow solid was shown by its ' H N M R and infrared spectra to be a mixture acid azide of 4-methylpentacyclo[4.2.O.O2~5.O3~8.O4~7]octanecarboxylic (18), ir 21 50 cm-' (N3), and 4-methyl-1-isocyanatopentacyclo[4.2.0.02~5.03~8.04~7]octane (19), ir 2260 cm-' ( N C O ) . Refluxing the mixture in benzene for 2 h gave complete conversion (ir) to the iso-
/ Proton Magnetic Resonance Spectra of Cubane Derivatices
3084 cyanate 19, obtained on evaporation as a yellow oil. 4-Methyl-N- chloride 29 began to precipitate within 5 min: yield 0.251 g (50%); mp 217 "Cdec. carbomethoxypentacyclo[4.2.O.O2~s.O3~8.O4~7]octylamine (20)was obPentacyclo(4.2.0.02~s.O3~8.O4~7~ctane_1,4-di~arbO~yli~ acid dimethyl tained by refluxing the isocyanate 19 for 12 h in methanol. Evaporaester (30)was prepared' I from diacid 22 as described above for comtion under vacuum gave a quantitative yield of 20 as a yellow solid, pound 11, and recrystallized from methanol, mp 164.5-166 OC (lit." mp 152-J58 OC, which sublimed to colorless needles, mp 158-159 "C. Pentacyclo[4.2.0.02~5.O3~8.04~7]octane-1,4-dicarboxylic Acid (22). 161-162 "C). A solution of ketal 1 ( 1 3.9 g, 0.0465 mol) in 75% w/w H2S04 (250 Pentacyclo[4.2.0.02~5.O3~8.04~7]octane-1,4-dicarboxylic acid dichloride (31)was prepared from diacid 22 as described above for ml) was stirred at room temperature for 3 days, then poured onto compound 12;mp 142-143 OC (lit.20 135-136 "C). crushed ice (500 g) and washed with dichloromethane (2 X 50 ml) Pentacyclo[4.2.0.02~5.O3~8.O4~7]octane-1,4-dicarboxylic acid dito remove unreacted 1. Continuous extraction of the aqueous phase teert-butyl perester (32)was prepared from the diacid dichloride 31, with dichloromethane for 5 days, followed by evaporation of the extracts, gave l-bromopentacyclo[4.3.O.O2~s,O3~8.04~7]nonan-9-one-4-terr-butyl hydroperoxide, and pyridine in dry ether, as described by Cole;20mp 133-137 OC (lit.20 136-137 "C). carboxylic acid (21)a s the hydrate, 10.4 g (82%), mp 216-220 O C (lit.Io 219-220 " C for the anhydrous ketone). Pentacyclo[4.2.0.02~s.03~8.04~7]octane (33)was prepared by pyrolysis of perester 32 in triisopropylbenzene at 150 OC as described by Cole.20 A solution of hydrate 21 (9.00 g, 0.0330 mol) in 25% aqueous soThe cubane 33 was purified by resublimation, mp 124-125 'C (lit.3 dium hydroxide (105 mi) was stirred under reflux a t l 15 'C for 2 h, 130-1 31 "C). then cooled and acidified to pH 3 by dropwise addition of concentrated 1,4-Dibromopentacyclo[4.2.O.O2~s.O3~8.O4~7]octane (34).Diacid 22 hydrochloric acid with stirring in an ice bath, maintaining the tem(0.926 g, 0.0048 mol) and mercuric oxide (2.4 g, 0.01 1 mol) were perature below 5 "C. (The dark brown solution turns light yellow at stirred in refluxing dibromomethane (50 mi). A solution of bromine the end point and a tan precipitate forms.) The solid was filtered off (2.1 g, 0.01 3 mol) in dibromomethane (25 ml) was added dropwise and recrystallized (glacial HOAc), to give 3.04 g (48%) of the diacid over 20 min and the mixture was refluxed overnight. The mixture was 22,mp dec above 225 "C (lit. 225,7b 226 "C"). filtered and evaporated to a solid which was extracted with hot hexPentacyclo[4.2.0.02~s.O3~8.O4~7]octane-1,4-dicarboxylic acid moanes. Evaporation gave the crude dibromide 34.0.91 g (72%), which nomethyl ester (23)was prepared by half-saponification of the diester was purified by recrystallization (hexanes) followed by sublimation, 30 in methanolic KOH a s has been described previously for the cormp 196- 198 OC. responding bicyclo[2.2.2]octane d e r i ~ a t i v e In . ~ a~ typical experiment, 1,4-Di(hydroxymethyl)pentacyclo[4.2.0.02~5.03~8.04~7]o~tane (35). 5.02 g of diester 30 (0.023 mol) and 1.28 g of potassium hydroxide (a) A solution of dicarboxylic acid 22 (0.292 g, 0.001 52 mol) in dry (0.023 mol) in I O : 1 methanol-water gave 1.89 g of recovered diester tetrahydrofuran (30 ml) was refluxed overnight with lithium alumi30 (38%), 0.98 g of diacid 22 (22%), and 1.80g of half-ester 23 (38%), mp 176-179 OC (benzene-hexane) (lit. 182-183,7b 174.5-176 OC9). num hydride (0.29 g, 0.0076 mol). The mixture was cooled and excess hydride decomposed by the addition of water (ca. 1 ml) and 15% Pentacyclo[4.2.0.02~5.03~8.O4~7]octane-1,4-dicarboxylic acid aqueous sodium hydroxide (1 ml). The mixture was then extracted monochloride monomethyl ester (24)was prepared by treatment of 23 with ether and the dried (MgS04) extracts were evaporated to a pale with thionyl chloride, as described above for 12;mp 109.5-1 1 I.5 "C. yellow solid. Recrystallization (chloroform-methanol) gave long white Recrystallization (hexanes) gave white needles of 24,in 96% yield, needles of dicarbinol35: 0.064 g (26%); m p 152-160 O C . mp 108.5-1 11.5 "C. 4-Hydroxymethylpentacyclo[4.2.O.O2~5.O3~8.O4~7]octanecarboxylic (b) A solution of ester acid chloride 24 (0.1 I8 g, 0.000 53 mol) and sodium borohydride (0.23 g, 0.0061 mol) in dry dioxane was stirred Acid Methyl Ester (25).To a cold solution of sodium borohydride for 8 h at room temperature and then poured onto ice. Extraction with (0.20 g, 0.0053 mol) in dry dioxane was added, with stirring, the acid chloroform, evaporation, and recrystallization (benzene-methanol) chloride 24 (0.100 g, 0.00045 mol). After I O min the mixture was gave the dicarbinol35,0.049 g (57%) as white needles, m p 164-165 poured into ice water (20 ml) and extracted with chloroform (4 X 20 OC,identical in spectral properties with the sample prepared in (a). ml). The organic phase was washed with ice water (2 X I O mi), dried (MgSO4), and evaporated. Recrystallization (hexane-benzene) gave Acknowledgment. We are grateful to the National Research 0.045 g of 25 (53%), mp 86-90 "C. Council of Canada for financial support, and to McGill UniPentacyclo[4.2.0.02~s.O3~8.O4~7]octane-l,4-dicarboxylic Acid Moversity for the award of a McConnell Fellowship (to G.E.L.). noamide Monomethyl Ester (26).The procedure followed was that We are also grateful to Mr. R. Simoneau for assistance in described by Roberts et al.s2 for the corresponding bicyclo[2.2.2]ocobtaining the spectra. tanes. Amide ester 26 was obtained in 48% yield, mp 232-237 " dec (ch!oroform-hexane) (lit.9 238-240 "C), together with a 26% recovery References and Notes of starting material, half-ester 23. (1)Pentacyc~0[4.2.0.0~5.03.8.04.~]octane. (2)P. E. Eaton and T. W. Cole, J. Am. Chem. Soc., 88, 962 (1964). Hofmann Rearrangement of the Amide Ester 26. Amide ester 26 (3)P. E. Eaton and T. W. Cole, J. Am. Chem. Soc., 88, 3157 (1964). ( I . I2 g, 0.0055 mol) was added in the dark to a solution of sodium (4)E. B. Fleischer, J. Am. Chem. Soc.,88, 3889 (1964). (0.26 g. 0.012 g-atom) i n dry methanol (20 ml). Bromine (1. I5 g, (5)B. D. Kybett, S. Carroll, P. Natalis, D. W. Bonnell. J. L. Margrave, and J. L. 0.007 mol) was added and the mixture was refluxed 20 min in the Franklin, J. Am. Chem. Soc.,88,626 (1966); J. L. Franklin and S. R. Carroll, ibid., 91,5940 (1969). dark, cooled, poured into water ( 1 75 mi), chilled overnight, and fil(6)J. C. Barborak. L. Watts, and R. Pettit, J. Am. Chem. Soc.. 88,1328 (1966). tered. A second crop of solid was obtained by extraction of the filtrate (7)(a) F. W. Baker, R. C. Parish, and L. M. Stock, J. Am. Chem. Soc.. 89,5677 (chloroform) and evaporation. The solids were shown by tic (10:l (1967);(b) T. W. Cole, Jr., C. J. Mayers, and L. M. Stock, ibid., 98,4555 chloroform-methanol on silica gel, sprayed with phosphomolybdic (1974). (8) L. Cassar, P. E. Eaton, and J. Halpern, J. Am. Chem. Soc.,92,3515,6367 acid and heated to develop the spots) to consist of three main com(1970). ponents, which were separated by fractional crystallization. These (9)L. J. Loeffler, S. F. Britcher, and W. Baumgarten, J. M e d . Chem.. 13,926 were: recovered starting material 26,0.235 g (21%) (from chloro(1970). (10)N. B. Chapman, J. M. Key, and K. J. Toyne, J. Org. Chem., 35,3860(1970). form); the expected urethane 4-(N-carbomethoxy)aminopentacy(11) T.-Y. Luh and L. M. Stock, J. Org. Chem.. 37,338 (1972). ~10[4.2.0.0~~~.0~~~.0~~~]0ctanecarboxylic acid methyl ester (27),0.090 (12)A. J. H. Klunder and 8. Zwanenburg, Tetrahedron, 28, 4131 (1972). g (7%). mp 147-1 50 "C (from CCL-petroleum ether 60-80 "C); and (13)A. J. H. Klunder and 8. Zwanenburg. Tetrahedron, 29, 1683 (1973). a dimeric acylurea N-(4-carbomethoxypentacyclo[4.2.0.0z~s. (14)T.-Y. Luh and L. M. Stock, J. Am. Chem. Soc., 98,3712 (1974). (15)J. D. Roberts and W. T. Moreland, J. Am. Chem. Soc., 75, 2167 (1953). 03~8.04~7]octanecarbonyl)-N'-(4-carbomethoxypentacy(16)L. M. Stock, J. Chem. Educ., 49,400 (1972),and references therein. cl0[4.2.0.0~~~.0~~~.04.7]octyl)urea(38),0.538 g (48%). mp 230-238 (17)In the notation of C. W. Haigh, J. Chem. Soc. A, 1682 (1970),"[AB], (C3J' " C dec (from DMSO). replaces the more cumbersome AA'A" BB'B" in denoting magnetically Curtius Rearrangementof the Azide Derived from 23.The procedure nonequivalent nuclei. We generally omit the point group symbol, as all [AB], systems with fixed configuration are either C3, or D3h and can be described above for 19 was used to convert 23 to 4-isocyanatopentaanalyzed as C3".lsb cyc10[4.2.0.0~~~.0~~~.0~~~]octanecarboxylic acid methyl ester (28):mp (18)For the theory and analysis of the limiting (AX13 case, see: (a) E. E. Wilson, 90-I I O "C; ir 2250 (NCO), 1720cm-' (COOMe). The crude product J. Chem. Phys., 27,60 (1957); (b) R. G.Jones, R. C. Hirst, and H. J. Bernwas hydrolyzed to 29 without further purification. stein, Can. J. Chem., 43,683 (1965). (c) See P. L. Corio. "Structure of High Resolution NMR Spectra", Academic Press, New York, N.Y., 1966.for a 4-Aminopentacyclo[4.2.0.02~s.03~s.04~7]octanecarboxylic Acid Hydiscussion of the general AA' . . . BB' . . . CC' . . . case. drochloride (29).Isocyanate 28,prepared from 0.50 g (0.00243 mol) (19)Other potential examples of [AB13 systems include 1,4disubstituted of half-ester 23,was refluxed for 30 min in a solution of hydrochloric barrelenes, and the two conformations of various hexasubstituted cycloacid (1.0 ml) in tetrahydrofuran (15 ml). The amino acid hydrohexanes, such as a//-cis-l,2,3,4,5,6-hexamethylcyclohexane (G. Mann
Journal of the American Chemical Society
/
98.1 I
/
May 26, 1976
3085 et al., Tetrahedron Left., 3563 (1970). (20) T. W. Cole, PhD. Thesis, University of Chicago, 1966. (21) A. A. Bothner-By and S. M. Castellano, "Computer Programs in Organic Chemistry", D. F. DeTar, Ed.. W. A. Benjamin, New York, N.Y. 1968, p 10. (22) The solvent shift effects and the assignment of shifts in aromatic solvents are discussed in part I1 of this series. (23) J. M. Key, Ph.D. Thesis, University of Hull, England, 1968. (24) A. J. H. Klunder, Ph.D. Thesis, University at Nijmegen, Netherlands, 1973. (25) G. L. Dunn and J. R. E. Hoover (S. K. F. Laboratories), British Patent 1 068 655;U.S. Patents 3 418 368; 3 538 160; 3 542 868; 3 562 317 (See Chem. Abstr., 68, p 2640m (1968)). (26) W. A. Gregory (E. I. du Pont de Nemours and Co.). U.S. Patent 3 558 704 (See Chem. Abstr.. 74, P 141105c (1971)). (27) E. S. Wallis and J. F. Lane, Oig. React., 3, 267 (1944). (28) I. Fleming and D. H. Williams, Tetraheon, 23, 2747 (1967). (29) (a) L. M. Jackman and S. Sternhell, Applications of Nuclear Magnetic Resonance Spectroscopy in Organic Chemistry", Pergamon Press, Oxford, 1969, p 280 ff; (b) p 164; (c) p 88. (30) (a) J. P. Schaefer and K. K. Walthers. Tetrahedron, 27, 5281 (1971); (b) S. Meiboom and L. C. Snyder, J. Chem. Phys., 52, 3857 (1970). and references cited therein; (c) for C-C-C and H-C-H angles of 90 and 112', respectively. (31) M. Karplus, J. Am. Chem. SOC.,85, 2870 (1963). (32) . . (a) . S. Sternhell, 0.Rev., Chem. SOC., 23, 236 (1969); (b) M. Barfield and B. Chakrabarti. Chem. Rev., 69, 757 (1969). (33) J. Hilton and L. H. Sutcliffe, Prog. Nucl. Magn. Reson. Spectrosc., 10, 27 (1975). (34) For an x-ray structure of cubane, see ref 4. The bond lengths and angles are essentially unchanged in 1,4dicarbomethoxycubane (P. Bird, private communication). (35) (a) M. Barfield and M. D. Johnston, Jr., Chem. Rev., 73,53 (1973); (b) s. L. Smith, Top. Curr. Chem., 27, 119 (1972).
(36) (a) R. W. Taft. E. Price, I. R. Fox, K. K. Andersen, and G. T. Davis, J. Am. Chem. Sac., 85, 709 (1963); (b) Taft et ai. do not report 61 values for COCl or NHC02R; as an approximation we have used their uIvalues for COF and NHCOR, respectively. (37) M. Charton. J. Org. Chem., 29, 1222 (1964); S. Ehrenson, R. T. C. Brownlee, and R. W. Taft, Prog. Phys. Org. Chem., 10, 1 (1973). (38) J. R. Cavanaugh and B. P. Dailey, J. Chem. Phys., 34, 1099 (1961). (39) P. R . Wells, Prog. Phys. Org. Chem., 6, 111 (1968). (40) M. T. Tribble and J. G. Traynham, "Advances in Linear Free Energy Relationships", N. B. Chapman and J. Shorter, Ed., Plenum Press, New York, N.Y., 1972, p 143, and references cited therein. (41) A. D. Cohen and T. Schaefer, Mol. Phys., 10, 209 (1966). (42) H. Booth and P. R. Thornburrow. Chem. lnd. (London), 685 (1968). (43) (a) D. J. Sardella, J. Mol. Spectrosc. 31, 70 (1969); (b) K. Takahashi. Bull. Chem. SOC.Jpn. 37, 291 (1968). (44) H. B. Evans, Jr.. A. R. Tarpley. and J. H. Goldstein, J. phys. Chem., 72, 2552 (1968). (45) R. C. Fort, Jr.. and P. v. R: Schleyer, J. Org. Chem., 30, 789 (1965). (46) K. B. Wiberg and V. 2 . Williams. Jr., J. Org. Chem., 35, 369 (1970). (47) T. M. Gund, P. v. R. Schleyer, G. D. Unruh, and G. J. Gieicher. J. Org. Chem., 39, 2995 (1974). (48) R. W. Taft and I. C. Lewis, unpublished summary quoted in ref 36a. (49) A partial analysis of the proton NMR spectrum of basketene has been reported (S.Masamune, H. Cuts, and M. G. Hogben, Tetrahedron Lett., 1017 (1966)).They observed meta couplings comparable to the values we have found in cubane (ca. 3.0 Hz), but surprisingly small ortho couplings (also ca. 3.0 Hz). No para coupling constants were measured. (50)Mass spectral fragmentations will be discussed in a subsequent paper. (51) A. J. H. Klunder and B. Zwanenburg, Tetrahedron, 29, 161 (1973). (52) J. D. Roberts, W. T. Moreland. and W. Frazer, J. Am. Chem. SOC.,75, 637 (1953); N. 8. Chapman, S. Sotheeswaran, and K. J. Toyne, J. Org. Chem., 35, 917 (1970).
Proton Magnetic Resonance Spectra of Cubane Derivatives, 11. Aromatic Solvent-Induced Shifts John T. Edward, Patrick G. Farrell,* and Gordon E. Langford Contributionfrom the Department of Chemistry, MrCill University, Montreal, Canada, H3C 3Gl. Received August 27, I975
Abstract: Aromatic solvent-induced shifts (ASIS) ( U C D C I ~ u ~ have ~ been ) measured for a number of substituted cubanes in benzene and pyridine and an additivity rule has been derived which allows accurate prediction of these shifts. Protons remote from the substituent show the largest ASIS and these are shown to correlate with substituent electronegativity. Models are discussed for the nature and stereochemistry of the solute-solvent interaction. It is suggested that the observed additive shifts arise from independent, transient 1:l associations of solvent molecules with the electron-deficient sites of all local dipoles i n the solute.
Stereospecific changes in chemical shifts induced by aromatic solvents have been reported for a wide variety of solutes, and a number of different models of the solute-solvent interaction have been proposed to explain In our studies of substituted cubanes, we found ASIS useful in removing accidental equivalences of chemical shifts when deceptively simple N M R spectra were obtained in CDC13. The observed shifts a r e highly stereospecific, are additive, and show interesting correlations with substituent electronegativities. In contrast to some other systems which have been studied, the rigid, fixed geometry of the cubane system makes it an ideal model for the investigation of both the nature and stereochemistry of these solute-solvent interactions. W e report here the results of our studies and confirm that the model proposed by Ronayne and Williams5 satisfactorily accounts for observed ASIS, both in this study and in those of other workers.
Results The compounds were synthesized, and their spectra were measured and analyzed using the computer program LAOCN3,6 in the manner ]previously described.' The observed chemical shifts and ASIS A = (VCDCI~ - V A ~for ) unsubstituted,
monosubstituted. and 1,4-disubstituted cubanes are shown in Table I; a positive A denotes a n upfield shift on replacing CDCl3 by the aromatic solvent. The observed solvent shifts with one exception' follow a consistent pattern which is summarized by eq I . A = So
+ (SI + 7 . 0 ) + S2
(1)
The parameter SOis the observed ASIS of unsubstituted cubane, i.e., 9.0 H z in benzene and 14.0 H z in pyridine. For disubstituted cubanes, S I and S2 are constants whose values depend both upon the substituents and their location (p, 7,or 6) relative to the proton whose shift is being calculated. Values of these specific substituent shift parameters, S,, are given in Table 11. For unsubstituted or monosubstituted cubanes, only the first one or two terms of eq 1 are used, respectively. Values of A calculated from eq 1 are included in Table I and agree to within f 3 H z with the observed shifts. In our previous paper1 we reported an additivity rule, eq 2 , = 403.8
+ Aul + Av2
(2) which accurately predicts chemical shifts of cubane and its mono- and 1,4-disubstituted derivatives in CDCls.8 Combining
Edward, Farrell, Langford
VCDIJ,
/
Aromatic Soluent Induced Shifts of Cubane Derivatives
